Evaluations and improvements in small area estimation methodologies

Small area estimation (SAE) of survey data down to small area level has become an increasingly widespread activity as scholars and policy-makers have sought to gain ever more detailed spatial information to better target interventions or resources and to evaluate local policy impacts. The availability of small area data has improved dramatically since the late 1990s yet many spatial variables of interest – income, fear of crime, health-related behaviours, and so the list goes on – remain impossible to access at small area geographies (i.e. beneath local authority level in the UK context). Various alternative methodologies have emerged to carry out SAE and these can be grouped broadly into statistical approaches and spatial microsimulation approaches, each with multiple differing approaches within them. A recent network, funded by the ESRC National Centre for Research Methods, brought together experts from across these methodological approaches and relevant external partners in order to enhance the state of the art in SAE through stimulating detailed comparative methodological discussion so as to better understand the strengths, weaknesses, similarities and differences between these methodologies. This methodological review paper emerges from the network discussions and aims to: summarise the main methodological approaches to SAE and their linkages; discuss the role of the small area covariate data and the opportunities and challenges around such data; identify the main methodological priorities around SAE in need of collective research attention; and, propose the need for a collective multi-methods SAE project in order more fully explore the conceptual and technical linkages between the statistical and spatial microsimulation methodologies